Prognostic indicators of survival for patients with oral cavity squamous cell carcinoma in Norway. Outcomes in a retrospective, multicenter cohort, with special focus on oral tongue squamous cell carcinoma, 2005-2009

(1)

The Faculty of Health Sciences Department of Medical Biology Tumor Biology Research Group

Prognostic indicators of survival for patients with oral cavity squamous cell carcinoma in Norway

Outcomes in a retrospective, multicenter cohort, with special focus on oral tongue squamous cell carcinoma, 2005-2009

Inger-Heidi Bjerkli

A dissertation for the degree of Philosophiae Doctor, May 2020

(2)

List of Tables

Table 1. ICD-10 diagnoses and anatomical sites of oral cavity cancer. ... 5

Table 2. TNM classification and stage for oral cavity cancer and the changes in the classification between 5th and 8th editions. Adapted from «TNM Atlas, 5th Edition» (23) and «TNM Classification of Malignant Tumours, 8th Edition» (22). ... 6

Table 3. Histopathological risk-parameters often presented in literature. ... 9

Table 4. The various variables in oral tongue squamous cell carcinomas considered in this thesis. .... 21

List of Figures

Figure 1 Estimates of head and neck cancers in Norway in 2008 and 2018. From "Cancer in Norway", 2008 and 2018 (with permission, and available from https://www.kreftregisteret.no/globalassets/) (7, 8)... 2

Figure 2 Oral cavity anatomy: a) oral tongue, buccal mucosa, and hard palate; b) floor of mouth, and gingiva and alveolar mucosa. Photographs: IH Bjerkli. ... 4

Figure 3 Flow-chart outlining the number of patients and available histopathological samples in the NOROC cohort presented in Papers I-IV. ... 17

Figure 4 Percentage and number of patients diagnosed with primary OCSCC at the four university hospitals in Norway, 2005-2009. ... 18

Figure 5 Tissue microarrays of oral tongue squamous cell carcinoma: a) overview H&E-stained TMAs; b) example of one pan-keratin stained TMA block (EA1EA3). Photographs: T Søland. ... 19 Figure 6 H&E-stained section of OTSCC with tumor budding; green arrowheads indicating buds<4, and black arrowheads indicating the tumor islands with>4 buds. Photographs: O Schreurs/T Søland. 20

(4)

Acknowledgements

The research for this thesis has been accomplished with the Tumor Biology Research Group (TBRG) in the Department of Medical Biology, Faculty of Health Sciences, University of Tromsø (Uit) — The Arctic University of Norway. This PhD-project was funded by grants from North-Norwegian Health Authorities (Helse Nord).

My work was carried out half time as a PhD-student, half time as a senior surgeon at the Department of Otorhinolaryngology at the University Hospital of North Norway (UNN), from 24 August 2014 to date. I sincerely thank Helse Nord for giving me this opportunity and for financing the PhD grant over the past years. This was done at the level of 50% over the six years, and in this way I was able to keep up with treatment of cancer patients half of my working time, and do research the other half.

All of the included patients in this study were previously diagnosed and treated at the four university hopitals in Norway dealing with head and neck cancers: the University Hospital of Oslo–

Rikshopitalet; Haukeland University Hospital, Bergen; St. Olavs University Hospital, Trondheim; and UNN, Tromsø. Without all of theses institutions and individuals, this work could not have been achieved.

Research is collaboration, and there are contributors I especially want to thank.

I thank everyone who has been part of the Norwegian oral cancer (NOROC) group from the four university hospitals in Norway on this project. Thank you for initiating, participating, discussing, and motivating in this long process over the recent years. Especially thanks to the other clinicians Olav Jetlund, Gunnhild Karevold, Ása Karlsdóttir, and Ellen A. A. Jaatun, for providing an office when I was collecting data at the other university hospitals. To all clinicians and pathologists: the

collaboration and especially our annual meetings have been an immense motivation to me!

I am especially thankful to my supervisors:

Principal supervisor Sonja Eriksson Steigen. I want to thank you for bringing me into the academic world, formulating the research questions, and giving me the opportunity to do research on a subject so close to my clinical work. You have been available, always responding, and a wonderful guide into the world of statistics and histopathology.

To my co-supervisors: Lars Uhlin-Hansen for formulating the research questions and the opportunity to do research in this group. Elin Hadler-Olsen, for valuable suggestions and always contibuting to the writing process of the research papers, no one responds as fast as you! Oddveig G. Rikardsen, for introducing me to the research group. All of you, and all the other co-authors, for valuable inputs and

(5)

discussions in the process of writing the papers. Especially co-author Tine S. Søland for your enthusiasm, academic thinking, and motivation for the second paper in this thesis.

Others in the TBRG, past and present, especially past and present team leader Gunnbjørg and

Synnøve, Gerd, Anna, Joffe, Rangita, Beate, Eli, Bente, Qalbi, Gaute, Kjersti, and Ann Iren S. You all gave me such an encouraging environment to work in at the university, and every week ending with a delicious Friday-cake event (apart from the last months of work in the lonely corona-lockdown situation)!

My second home, the Department of Otolaryngology at UNN, who granted me a leave of absence. A great thank you to the head of my department, Torgrim Fuhr, for believing in my research, and to all the other college surgeons past and present at the Department of Otorhinolaryngology, especially the Division of Head and Neck. Especially thank you, Ann Iren H, with whom I share an office.

My best friends; Tone, Annette, Myrdene, Åshild, Merethe, Gunnhild, Ellen, Nadia, Anne, Julie, Janet, Mats and Line for always supporting and listening to me. Especially thanks for essential help in translations and linguistic questions to Myrdene. I think of you all as my extended family!

To my dear family! Especially to my parents, and indeed to my father, who during my years of PhD- work, unexpectedly became one of the men in oral cavity cancer statistics. The natural fear that our younger child well adressed by asking, “Will grandpa be able to speak and tell stories again, or will he die?” — gave me extra motivation to fulfill this work and the continuous work for this group of patients. To my brothers, especially my youngest brother Pål, often on the phone, giving support. To my cousins, John-Erik and Eva, for always listening, believing, and supporting me since childhood.

And to my dear husband Børge; the way you have supported me all the way through is of paramount importance to me, I truly admire your patience! To our precious children; Klara and Elmer, you are my true guiding stars in life, and I look forward to spending so much more time with you!

Tromsø, May 2020 Inger-Heidi Bjerkli

(6)

(7)

Abbreviations

AJCC American Joint Committee on Cancer AKF The Unit of Applied Clinical Research CRF Case report form

CRN The Cancer Registry of Norway

DAHANCA The Danish Head and Neck Cancer Group DOI Depth of tumor invasion

DSS Disease-specific survival

ECOG Eastern Cooperative Oncology Group EHR Electronic health record

ESAS-r Edmonton Symptom Assessment System-revised version FNAC Fine needle aspiration cytology

H&E Hematoxylin and eosin HN Head and neck

HPV Human papillomavirus

ICD-10 International Statistical Classification of Diseases and Related Health Problems 10th Revision

IHC Immunohistochemistry IRB Institutional Review Board ISH In situ hybridization MDT Multidisiplinary team

NCCN National Comprehensive Cancer Network NOROC NORwegian Oral Cancer study

OCC Oral cavity cancer

OCSCC Oral cavity squamous cell carcinoma OS Overall survival

OTSCC Oral tongue squamous cell carcinoma

REK The Regional Committee for Medical Health Research Ethics REMARK REporting recommendations for tumor MARKer prognostic studies RT Radiation therapy

SCC Squamous cell carcinoma

SEER The Surveillance, Epidemiology and End Results Program SNOMED Systematized NOmenclature of MEDicine

SPSS Statistical Package for the Social Sciences

TB Tumor budding

TMA Tissue microarray TNM Tumor-Node-Metastasis cTNM clinical presentation TNM pTNM pathological presentation TNM

UICC Union for International Cancer Control UNN University Hospital of North Norway WHO World Health Organization

WPOI Worst pattern of invasion

(8)

List of papers

This thesis is based on the following papers. They are referred to as Papers I-IV in the manuscript.

Paper I

Characteristics and prognosis of primary treatment-naïve oral cavity squamous cell carcinoma in Norway, a descriptive retrospective study. Bjerkli IH, Jetlund O, Karevold G, Karlsdóttir Á, Jaatun E, Uhlin-Hansen L, Rikardsen OG, Hadler-Olsen E, Steigen SE.

PLoS One. 2020 Jan 16;15(1): e0227738. doi: 10.1371/journal.pone.0227738. eCollection 2020.

Paper II

High-risk human papilloma virus was not detected in a Norwegian cohort of oral squamous cell carcinoma of the mobile tongue. Søland TM, Bjerkli IH, Georgsen JB, Schreurs O, Jebsen P, Laurvik H, Sapkota D. Manuscript submitted for publication.

Paper III

Tumor budding score predicts lymph node status in oral tongue squamous cell carcinoma and should be included in the pathology report. Bjerkli IH, Laurvik H, Nginamau ES, Søland TM, Costea D, Hov H, Uhlin-Hansen L, Hadler-Olsen E, Steigen SE.

Manuscript submittet for publication.

Paper IV

A combined histo-score based on tumor differentiation and lymphocytic infiltrate is a robust prognostic marker for mobile tongue cancer. Bjerkli IH, Hadler-Olsen E,

Nginamau ES, Laurvik H, Søland TM, Costea D, Uhlin-Hansen L, Steigen SE. Manuscript submitted for publication.

(9)

Abstract

Background: Oral cavity cancer (OCC) is the most frequent of all head and neck (HN) cancers, and has distinct survival outcomes compared to other sites. The oral tongue is the most common site of cancer in the oral cavity. About 90% of these are squamous cell

carcinomas (SCC), and they can be very aggressive with a high mortality rate. The treatment of these cancers is based on the tumor (T), lymph node (N), and metastasis (M) classification, although tumors with the same TNM classification may act differently in aggressiveness.

Treatment of oral cavity (OC)SCC is preferentially surgery when the tumor is regarded as resectable, with additional neck-dissection for many of them. For some patients, postsurgical radiotherapy (RT) is added, and chemotherapy can be used in a palliative setting. Today there are no established histopathological or molecular markers in use to differentiate between those who will benefit from additional neck-dissection, or RT, when no lymph node metastases primarily are suspected. Such additional markers would be useful for

supplementing the commonly used TNM classification in treatment decisions. Objective: The goal of this PhD-project was to collect clinical and histopathological information for a

national cohort of OCSCC in the period 2005-2009, the Norwegian oral cancer (NOROC) study, to assess survival and prognostic factors. We explored whether high-risk Human Papilloma Virus (HPV) was present in oral tongue (OT)SCC, and assessed the prognostic value of different tumor growth patterns of these cancers. Methods: Clinical information was retrieved from the patients’ electronic hospital files. Histologic sections were reexamined.

Data from the national Cause of Death Registry was used to calculate overall and disease- specific survival. Statistical evaluation was performed to determine correlations, and find independent and significant predictors of survival. Results: We identified 643 patients with OCC, 535 of these were primary treatment-naïve OCSCC. Age at time of diagnosis, and low- stage disease correlated with higher survival outcome. We did not detect high-risk Human Papilloma Virus in the OTSCC. Tumor depth of invasion shifted many of the tumors to a higher T-status. Tumor differentiation, tumor budding, and lymphocytic infiltrate were the most important histopathologic prognosticators. Conclusions: Histopathological variables such as tumor budding, tumor differentiation, and lymphocytic infiltration, can add significant prognostic information to aid clinicians in treatment decisions and follow up, especially for low-stage tumors.

(10)

Sammendrag

Bakgrunn: Munnhulekreft er den hyppigst forekommende kreftformen i hode- og halsområdet. Munnhulekreft kan vokse aggressivt, og har lav overlevelsesprosent, nesten halvparten av pasientene dør i løpet av de første fem år etter behandling. Behandlingen avgjøres basert på pasientens generelle helsetilstand, men mest av alt med bakgrunn i

klassifiseringen av kreftsykdommen, en såkalt TNM-klassifisering. T står for svulststørrelse, N for spredning til lokale lymfeknuter og M for fjernspredning. Sykdom med samme

klassifisering kan ha forskjellig forløp. Munnhulekreft behandles først og fremst ved kirurgi, i tillegg kan det gis strålebehandling og i sjeldnere tilfeller kjemoterapi. Vi har så langt ingen veletablerte histopatologiske eller molekylære markører til bruk for å differensiere mellom behandlingsmodaliteter for munnhulekreft. Målsetting: Vårt mål var å etablere en nasjonal kohort av munnhulekreft. Vi reklassifiserte patologiske vevsprøver fra pasienter med kreft i den mobile tunge. Vi ønsket å undersøke om høyrisiko humant papillomavirus (HPV) er tilstede i tungekreft. Vi ønsket også å undersøke om ulike aspekter ved vekstmønster i disse kreftsvulstene er av prognostisk verdi. Metode: Dette doktorgradsprosjektet er basert på en retrospektiv innsamling av datamateriale fra pasienter som fikk påvist og behandlet

munnhulekreft i Norge i årene 2005-2009, og reklassifisering av tilgjengelige

histopatologiske prøver. Studien fikk navnet NOROC, Norwegian Oral Cancer study. Vi innhentet data fra Dødsårsaksregisteret for å kunne beregne overlevelse. Forskjellige statistiske metoder ble brukt for å undersøke sammenheng mellom variabler og

overlevelsesanalyser. Resultat: Vi har beskrevet en stor kohorte med munnhulekreft fra Norge, og fant 643 tilfeller der munnhulekreft var satt med ICD-10-diagnose. 535 av disse var førstegangs tilfeller av kreft i munnhulen. Median alder for diagnosen var 67 år, fem års total overlevelse var 47%, fem års sykdoms-spesifikk overlevelse var 52%. Vi fant at yngre alder, små svulster med lav sykdomsutbredelse (N0) korrelerte med høyere overlevelse. Vi fant ikke høyrisiko HPV i vårt materiale av tungekreft. For vekstmønster fant vi at dybdevekst forskjøv mange av tumorene til høyere klassifisering. Tumor budding, differensieringsgrad sammen med lymfocytt infiltrat, de siste to i en kombinert histo-skår, kan være med som supplement til TNM-klassifiseringen. Konklusjon: Vi har demonstrert at histopatologiske variabler som tumor budding, differensieringsgrad og lymfocytt infiltrat, kan tilføre viktig informasjon.

Dette kan brukes til å predikere for aggressivitet i tumor og bidra sammen med TNM- klassifiseringen i omfanget av behandling og videre oppfølging.

(11)

(12)

1 Introduction

Oral cavity cancer (OCC) is one of the most common subsites of head and neck (HN) cancers.

In all treatment of cancer, the patients should be able to know what disease they are facing, by receiving current and correct information. Clinicians and patients need to decide upon

treatment based on the most updated knowledge and most likely outcome, and the patient has to provide consent for treatment from this perspective. Consequently, there is a need for continuous research in epidemiology, diagnosis, management protocols, and outcomes for all forms of cancer. In this way, we are able to make decisions according to the principles of evidence-based medicine.

1.1 Epidemiology

HN cancers constitute 2-5% of all cancers in the world (1). Lip and OCC, when joined together, are the most common subtypes of HN cancers, and these comprised 354,864 new cases worldwide in 2018, 2% of all cancers. Oropharyngeal (oral pharynx) cancer comprised 92,887 new cases the same year, 0.5% of all cancers (1). These subsites should not be merged as they are different entities.

In 2012, the global incidence of oral cancer was estimated at 275,000 per year, and the incidence is steadily rising worldwide, according to global cancer statistics in 2018. Lip and OCC rank as the 18th most common of all cancers with 354,864 new cases, and rank as the 16th most frequent of deadly cancers in the world, with 177,384 deaths in 2018 (1). In general, men are at higher risk for having OCC than women (1). Median age for OCC is in the mid-sixties, but for oropharyngeal cancer, especially the human papillomavirus (HPV)- positive oropharyngeal cancer, the median age is ten years younger (2-4). Some studies report patients with oral tongue cancer to be younger than patients with other cancers of the oral cavity (5, 6).

The numbers of HN cancers have been rising in Norway, as well as globally (1, 7, 8). In the latest annual surveys from the Cancer Registry of Norway (CRN), OCC has been recorded

(13)

cancers are classified to their anatomical location according to the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) diagnosis (9). Previously the national cancer registry tended to merge oral cavity and oropharyngeal sites; the number of new cases by primary sites and sex, from the CRN, for 2008 and 2018, is shown in Figure 1 (7, 8).

Figure 1 Estimates of head and neck cancers in Norway in 2008 and 2018. From "Cancer in Norway", 2008 and 2018 (with permission, and available from

https://www.kreftregisteret.no/globalassets/) (7, 8).

In Norway, the total number of patients with new cases of OCC (ICD-10; C02-C06) was 181 in 2008 (although this number includes C01, an oropharyngeal site), and 244 in 2018 (7, 8).

The oral tongue is the most common site for OCC, comprising up to 50% of the cases (2, 10, 11).

Cancers occurring in different anatomical sites often have distinct etiologies and different treatment options, resulting in different survival outcomes. HN cancers are often presented and discussed as one type of cancer although the term includes different entities of cancer (12). It is important to describe precisely the sub-cohort of HN cancer patients when

(14)

estimating the most probable outcome. Oral cavity cancer and oropharyngeal cancer are two distinct cancer sites with different etiology, treatment, and survival outcome (13). In earlier investigations, data from oral tongue cancer (oral cavity site) and base of tongue cancers (oropharyngeal site) have often been combined and called tongue cancers or oral cancers, without being distinguished, and these studies are hampered by bias with respect to both risk factors and survival outcomes (14-16). In other reports, OCC incidence is presented together with cancer in the lip, and global cancer statistics combine these two locations in their

presentations (1). Lip cancers tend to act like non-melanoma skin cancers and are for the most part less aggressive than cancers of the oral cavity (1, 17, 18). As a consequence, it is

important to describe cohorts with validated oral cavity sites that are not merged with other HN sites in order to have reliable information.

In most countries, studies on patients with OCC report a five-year survival rate around 50%

(4, 12, 18). In 2019, the Surveillance, Epidemiology, and End Results (SEER) program

database in the U.S., published a five-year relative survival rate of 53% for the floor of mouth, and 66% for tongue, for the years 2009-2015 (19). It is difficult to compare different reports, as many merge anatomical sites, or do not specify precisely the anatomical sites they are describing, and there is no consistency in reporting survival as relative survival, overall survival (OS), or disease-specific survival (DSS). In some cases, differences in reporting cancer cohorts occur regardless of treatment options, and some studies report survival of the patients treated in curative intent, and exclude those with metastasized cancer, and this will also preclude any proper comparison of cohorts (2, 20). The anatomical regions are nowadays better clarified by the World Health Organization (WHO), and one should describe cohorts according to this resource (21).

Squamous cell carcinomas (SCC) are the most common malignancy in the oral cavity, representing 90% of the cases. Less common are verrucous carcinomas, adenocarcinomas, melanomas, lymphomas, and sarcomas (21).

In this thesis, we present the Norwegian Oral Cancer (NOROC) study, in which we focus on SCC of the oral cavity, with both clinical data and histopathological data, documenting diagnosis, treatment choices, and follow-up. All this in order to present a homogeneous cohort concerning the ICD-10 diagnosis, one type of cancer cells, and survival outcome.

(15)

1.2 Anatomy and classification of tumors

The oral cavity consists of the oral tongue, floor of the mouth, hard palate, buccal mucosa including the retromolar areas, and upper and lower alveolus and gingiva (gum), as shown in Figure 2 a-b. With respect to discussing locations and management of OCC, the designation

“oral tongue” is applied to the mobile/anterior two-thirds of the tongue. The remaining posterior one-third is the base of the tongue and considered as part of the oropharynx (22).

Figure 2 Oral cavity anatomy: a) oral tongue, buccal mucosa, and hard palate; b) floor of mouth, and gingiva and alveolar mucosa. Photographs: IH Bjerkli.

Tumors and cancers are classified by anatomical location according to ICD-10 with an alphanumeric code, possibly with an eventual fourth character. ICD-10 diagnoses applied to cancer in oral cavity are C02-C06 (9), but do not include C05.1 (soft palate) and C05.2 (uvula), as these are regarded oropharyngeal sites. C02.4 (tongue-tonsils) should also be exclusively used for oropharyngeal cancers (22). The ICD-10 diagnoses for OCC are shown in Table 1.

(16)

Table 1. ICD-10 diagnoses and anatomical sites of oral cavity cancer.

ICD-10 diagnosis Anatomical site of oral cavity cancer

C02 Oral part of tongue

C03 Upper and lower gum

C04 Floor of mouth

C05 Hard palate

C06 Buccal mucosa

The anatomical extent of disease is classified in the TNM system. The TNM system describes the extent of the primary tumor (T); for head and neck cancers, the extent (absence or

presence) of cervical (regional) lymph node metastasis (N); and the absence or presence of distant metastases (M). The TNM classification and staging of disease (I-IV) are described according to TNM Union for International Cancer Control (IUCC) and American Joint Committee on Cancer (AJCC) (17, 21-23). The simplicity of the TNM system promotes clinical utility, but the prognostic outcome can be difficult to determine in patients with oral cavity squamous cell carcinoma (OCSCC) (24).

The TNM system has been revised on an irregular basis, but in 2017 a new classification was introduced (22, 24). In the new TNM 8thedition, for OCC, tumor dimension now including depth of invasion (DOI), is a standard assessment in the T classification. Cervical lymph nodes surgically removed will be described with or without extranodal extension (22). For the oropharyngeal cancers, HPV-status was included. The categories in the former TNM in use for the years 2005-2009, and those in the new 8th edition of TNM classification for OCC, are shown in Table 2.

(17)

Table 2. TNM classification and stage for oral cavity cancer and the changes in the

classification between 5th and 8th editions. Adapted from «TNM Atlas, 5th Edition» (23) and

«TNM Classification of Malignant Tumours, 8th Edition» (22).

TNM 5th edition TNM 8th edition (if changed)

Primary tumor (T) classification, size and extent

TX Primary tumor cannot be assessed

T0 No evidence of primary tumor

T1 Size ≤2 cm in greatest dimension and 5mm or less in depth of invasion (DOI) T2 Size > 2cm<4 cm in greatest dimension Size≤2cm and DOI 5-10mm, or size >2

cm<4cm and DOI≤10mm T3 Size >4 cm in greatest dimension or DOI >10 mm

T4a/b Tumor invades adjacent tissue

Cervical node (N) metastasis classification

NX Regional lymph nodes cannot be assessed N0 No regional lymph node assessment

N1 Metastsis in a single ipsilateral node, and without extranodal extension

≤3 cm in greatest dimension

N2 a) Metastasis in a single ipsilateral>3 cm< 6cm a) and no extranodal extension b) Ipsilateral, multiple ≤6 cm b) and no extranodal extension c) Bilateral, contralateral≤6cm c) and no extranodal extension

N3 >6 cm a) and no extranodal extension

b) Any N with extranodal extension Distant metastasis (M) classification

MX Distant metastasis cannot be assessed considered to be inappropriate to use

M0 No distant metastasis

M1 Distant metastasis

Stage of disease on basis of TNM classification

Stage I T1 N0 M0

Stage II T2 N0 M0

Stage III T1-T2 and N1 M0

T3 and N0-N1 M0

Stage

IVA T1-T3 N2 M0

T4a N0-2 M0

Stage

IVB Any T N3 M0

T4b any N M0

Stage

IVC Any T any N M1

(18)

The use of the TNM classification system guides in the choice of treatment and provides a rough prediction of probable outcome. Moreover, the TNM system of classification facilitates treatment evaluation and the exchange of information between clinicians, supports cancer control activities, and contributes to standardize cancer research. For these reasons, it is important that the classification system remains stable, in order to evaluate factors contributing to long-term survival outcomes in a population (22).

1.3 Etiology of oral cavity cancer

Globally, there are large geographical variations in incidence of HN and cancer of the oral cavity (1). Cultural habits, such as betel nut quid chewing and eating areca nuts, increase the OCC incidence in South and South East Asia (4, 25, 26). Cigarette (tobacco) smoking and excessive alcohol consumption are major risk factors (27-29) . We also know there is a synergistic effect of smoking and alcohol consumption (28). Poor dental health status has also been considered to be a risk factor, although reports are inconsistent (30, 31). On the other hand, other factors like consuming vegetables and fruits might forestall development of cancer (29, 32).

Viruses (oncoviruses) play a role in some HN cancers; especially documented is the Epstein- Barr virus in nasopharyngeal cancer, and HPV in laryngeal and oropharyngeal cancers (17).

In the last decade, high-risk HPV has been established as playing an important role in oropharyngeal cancers, and this has therefore been implemented in the new TNM classification (22). Some studies have reported a small proportion of OCC presenting as HPV-positive (33, 34). HPV detection and confirmation can be done by several methods (35, 36). We lack larger studies on high-risk HPV in OCC where the anatomical locations for the tumors are precisely described.

The risk of recurrence or second primary tumors is present for the patients (37, 38). Genetic factors influence the initiation and progression of cancers, but the molecular mechanisms remain uncertain; several biological concepts have been described by Hanahan and colleagues in «The Hallmarks of Cancer» in 2011 (39).

(19)

1.4 Diagnosing oral cavity cancer

1.4.1 Clinical and radiological considerations

Early detection is important to improve the likelihood of good survival outcomes (2, 12, 40).

The most common clinical manifestations of OCC are pain or numbness, wounds not healing, dental-symptoms, and cervical lymph node presentation (palpable lumps or mass) (21).

Patients can present with their signs and symptoms to the general practitioner, an ear-nose- throat /HN-specialist, or a dentist or dental hygienist. In Norway, patients with suspected HN cancer are referred to one of the four university clinics in the country, either Oslo University Hospital–Rikshospitalet in Oslo, Haukeland University Hospital in Bergen, St. Olavs

University Hospital in Trondheim, or University Hospital of North Norway (UNN), Tromsø.

The patients are clinically examined, documented byradiological imaging such as ultrasound, computed tomography (CT) scan, orthopantomogram (OPG), and sometimes magnetic

resonance imaging (MRI) and positron emission tomography (PET)-CT (41, 42). CT is the most-used imaging modality for OCSCC in the Nordic countries (43). The clinical and radiological appearance forms the basis for the clinical (c)TNM classification of cancer disease.

1.4.2 Histopathology

Nearly 90% of the cancers of the oral cavity are SCC, originating from within the epithelial (mucosal) lining (21). The SCC in the HN region is heterogenous in histopathologic

presentation, clinical appearance, and response to treatment. Samples, such as biopsies, and fine needle aspiration cytology (FNAC), are examined histologically for signs of pathology by a clinical pathologist. The histopathological verification forms the pathological (p)TNM classification (22). Histopathological confirmation of the clinico-radiological diagnosis is the foundation for tumor classification and further treatment decisions.

OCSCC are further classified according to morphological evaluation, although grading alone does not correlate well with prognosis (21). WHO distinguishes three morphological grades

(20)

divided into well, moderately, and poorly differentiated (grade I-III). Within a tumor, the degree of differentiation varies as well (21).

Within the same TNM classification, individual differences in clinical outcome is present, especially for the lower stages. Low-stages (early-stages) tumors are stage I and stage II (T1N0M0 and T2N0M0, in TNM 8thedition, with tumor size <4cm and/or DOI ≤ 10 mm, with no lymph node or distant metastasis) of OCSCC (44).

Tumor budding (TB) has been proposed as a possible additional variable for colorectal cancer (22, 45), and some have argued that this characteristic may also be of value for appraising OCSCC evaluation (46, 47). In fact, before DOI was introduced into the new TNM classification, some studies also suggested applying a combined TB and DOI score (BD score) as a prognostic indicator (48, 49).

Guidelines for recognizing patterns of the invasive front of the tumor and histologic risk models have been proposed by Anneroth and coworkers in 1987, Bryne and coworkers in 1998, and by Brandwein-Gensler and coworkers in 2005 (50-52). These latter classifications have been tried in later studies, but with only moderate success, as differentiating high-risk tumors and low-risk tumors is very challenging (44, 53-55). Proposed histopathological risk- parameters of tumors are: WHO differentiation grade, worst pattern of invasion (WPOI), degree of keratinization, nuclear polymorphism, perineural invasion, lymphocyte infiltration, and vascular infiltration (17, 21, 50-52). Some of the different histopathological growth pattern assessments presented in literature over recent years are shown in Table 3.

Table 3. Histopathological risk-parameters often presented in literature.

Histopathological variables References

Tumor depth of invasion (DOI) (22)

Tumor budding (TB) (45, 46)

Tumor budding and depth of invasion score (48)

WHO differentiation, whole tumor (17, 21)

WHO differentiation, worst pattern (52)

Degree of keratinization, whole tumor (50)

Degree of keratinization, tumor front (51)

Nuclear polymorphism, whole tumor (50)

Nuclear polymorphism, tumor front (51)

Perineural infiltration (52)

Lymphocytic infiltrate (52)

(21)

Several studies have tried to adapt to these oral cancer risk-parameters in histopathology.

However, few studies have been large enough, or based strictly on one anatomical site, to allow conclusive results. Different pathologists notice different patterns (inter-rater) and they will not always agree with themselves (intra-rater). The reproducibility between pathologists has potential for improvement with respect to inter-rater and intra-rater agreement (53, 56).

1.5 Treatment options and treatment complications

Multidisciplinary team (MDT) meetings are central in all cancer treatment and care (57, 58).

The treatment options of today require accurate tumor classification and staging as well as additional molecular details. As a consequence, an increasing number of procedures for diagnosing and staging of cancer in patients is needed to determine any treatment decision (59). Today, treatment requires involvement of experts within different medical disciplines, and specialists from these disciplines meet in MDT.

For many cancers, a range of innovative surgical and medical treatments have been developed over recent decades. Examples include minimal invasive surgery techniques, stereotactic radiation therapy (RT), numerous chemotherapeutics, and involvement of targeting therapies and immune-therapies, all broadening the choice of treatment for most cancer types (60).

For OCC, the most common choice of treatment is still surgery of the primary site when the tumor is regarded resectable, with or without surgical neck dissection (61). Additional postoperative RT may be given (62). In some of the cases, RT is given before surgery to decrease the size and extent of the tumor (59, 63-65). A more advanced stage of disease (≥stage III) will, for some patients, lead to additional medical treatment with chemotherapy or combined radiochemotherapy. Chemotherapy alone is for the most part used for metastasizing disease and in a palliative setting (66, 67).The treatment in many cases leads to some sequela, depending on the size of the tumor and the extent to which the patient has had neck dissection surgery, reconstructive surgery, and postoperative RT.

As many as one-fifth of patients with low-stage tumors might experience neck node

metastasis and recurrence within 1-3 years (68, 69). Two options are discussed for treatment of the neck in clinically node-negative (N0) OCSCC. One is elective neck dissection, and the

(22)

other is wait-and-see. In some cases for low-stage diseases, clinicians might suggest watchful waiting rather than neck dissection surgery (70-72). A patient with a small tumor and a N0- status in the neck region does not necessarily need neck dissection or postoperative RT, but early-stage node-negative OCSCC, that later develops regional metastasis, has poor prognosis (70, 71).

Complications of OCC treatment are associated with physical and psychological debility.

Some patients feel socially inhibited due to a changed facial appearance and disfiguring scars, or because of persistent speech, swallowing, or tracheostomy difficulties. Xerostomia is common and can cause poorer dental health. Alterations in taste, malnutrition, and trismus are other complications. Moreover, a late complication of RT is induced osteoradionecrosis of the mandible, and some report an increased risk of ischemic stroke many years after RT (73). The patients may perceive themselves as altered, and many are not able to work again, part-time or full-time. Any impact on quality of life will be a cost for the individual as well as for society (74-76).

There is no established tumor growth pattern or biomarker profile to identify patients with occult neck metastasis that would allow for more individualized treatment, especially for the patients with low-stage disease (77). If such a diagnostic tool were available, we could predict aggressiveness of the tumor and limit treatment-related adverse effects. Due to the severe side-effects of the treatment, it is important to reduce overtreatment of OCC patients. In the absence of a national guideline for treatment of HN cancers in Norway, the Danish Head and Neck Cancer Group (DAHANCA) and the U.S. National Comprehensive Cancer Network (NCCN) guidelines have been used (59, 63).

Many previous studies that describe treatment and survival outcome of patients with OCC have not been consistent in reporting the anatomy and ICD-10 diagnosis or the actual

treatment given. Some studies only included a small number of patients, and in some studies different anatomical sites have been merged (12, 78). In several studies, SCC of the

oropharynx may be presented as part of the oral cavity and vice versa (13, 16). This predicament makes discussions of etiology and survival outcome inconclusive. To better understand OCC and answer our own and our patients’ questions, it was considered important to describe a cohort of solely primary OCC, that is, squamous cell carcinomas, since they make up 90% of oral cancer patients. We wanted to provide a more precise description of epidemiology, etiology, treatment, and survival outcome in a national cohort that is

(23)

comparable to other studies. Only in this manner can we describe and argue for the choice of treatment and give prognostic outcomes to both our patients and our colleagues. A transparent cohort would be valuable for future studies to adhere to the Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK) guidelines (79-81).

(24)

2 Aims of the thesis

2.1 Purpose of the thesis in general

The general aim was to establish a large cohort of oral cavity squamous cell carcinomas with a broad overview concerning clinical and histopathological characteristics. All this in order to find characteristics with prognostic value. The specific aims of my PhD-work were:

1. To establish a national, multicenter cohort of patients strictly limited to OCSCC, excluding other HN cancer sites;

2. To describe this cohort with respect to clinical and histopathological characteristics, and to calculate survival data;

3. To determine if high-risk HPV is present in oral tongue (OT) SCC;

4. To assess prognostic value of various morphological tumor traits that have previously been suggested as prognostic markers for OCSCC.

Our contributions fulfilling the stated aim of this thesis are presented in four separate articles, each with different research aims.

2.2 Aims of the included papers

• In Paper I we wanted to characterize a retrospective, multicenter cohort of solely OCSCC from all four health regions in Norway. We aimed to find clinical

characteristics that significantly influenced survival, which we could compare with other cohorts.

• In Paper II we aimed to determine if high-risk HPV was found in a cohort with OTSCC.

• In Paper III we wanted to explore how depth of tumor invasion and tumor budding in OTSCC can impact survival outcomes.

• In Paper IV we aimed to describe how different high-risk histopathological patterns, even when made less complex with fewer options, can predict survival outcomes in OTSCC.

(25)

(26)

3 Materials and methods

This section provides an overview of the patient material retrieved retrospectively as the basis for the NOROC study, a collaborative multi-center study in Norway, outlining methods used.

In Norway, treatment of HN cancer is centralized to the four university hospitals. The Institutional Review Board (IRB) of the Northern Norwegian Regional Committee for Medical Research Ethics (REK Nord) approved the NOROC study in 2013, with an

expansion in 2015, and an extension in 2019. The protocol numbers are REK Nord 2013/1786 and 2015/1381 (Appendices I, II, III and IV). Ethical and methodological considerations are summarized in chapter 5.

3.1 The web-based case report form

The main build-up of the case report form (CRF), was developed in cooperation with the clinicians and pathologists of the NOROC group. Data collection was enabled by a web-based data collection system developed and administered by The Unit of Applied Clinical Research (AKF), Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology, Trondheim, Norway (82). The web-based CRF was administered on different servers, all situated in Norway. To be able to record, the person had to log in with a personal ID and a two-step message-access control code (double-authentication). Each of the four university hospitals had one clinician and one pathologist with access to record data into the CRF, a total of eight persons. The vast majority of the recording of clinical data was made by the PhD-candidate (IHB) as a clinician, supplemented by the other clinicians at the other hospitals. Four different pathologists were responsible for the histopathological re-assessment and recording. A list of questionnaire-specific questions and answers is included in Appendix V.

(27)

3.2 Data collecting process

3.2.1 Extracting clinical data from the patient cohort

Electronic health records (EHR) have been introduced in hospitals in many countries and, for Norwegian hospitals, from the beginning of 2000. Digitalized journals were implemented in Norway from 2002, but the latter were only fully available from 2005. After the introduction of EHR, we could search for the diagnoses electronically, and we did not have to find and store large collections of health records on paper (83). The time span 2005–2009 was chosen for data collection because we wanted to have a minimum of 500 patients and to have a minimum of a five-year follow-up in 2015; hence the last year for diagnosis had to be 2009.

The collection of clinical data was achieved between August 2015 and February 2017.

The inclusion criteria were that patients had to be diagnosed with OCSCC (ICD-10 C02-C06) in the time span 1 January 2005–31 December 2009, documented in the EHRs. Exclusion criteria were: any reservation about research recorded in the EHR, the cancer not being in the oral cavity but in a neighboring site (like C05.1, C05.2, or in the tongue-tonsil C02.4), cancers other than SCC, recurrent cancer or second primary from a cancer before 2005, or any record of previous cancer treatment.

The ICD-10 diagnoses were matched with the pathologists’ archive for patient tumors with coding T51 (mouth) and T53 (tongue). These are codes in the pathologists’ archive called the Systematized Nomenclature of Medicine (SNOMED) (84). Some patients did not have the correct diagnosis in the EHR, and some were registered twice with different codes within the oral cavity or twice in different hospitals. We found 643 patients with OCC (ICD-10; C02- C06). The NOROC study focused on SCC, so therefore we excluded other histopathological diagnoses such as verrucous carcinoma, malignant melanoma, adenoid cystic carcinoma, lymphomas, and sarcomas. When all diagnoses, clinical data, and available histopathological samples were assessed, we ended up with 535 unique primary treatment-naïve OCSCC, as described in Paper I. When we extracted the strictly oral tongue cancers from the original 643, we found a total of 273 patients that had OTSCC; 34 of these were recurrences or second primaries. Those of the 273 patients with OTSCC available for tissue microarray (TMA) technique (n=146), formed the cohort for Paper II. When focusing on the primary treatment- naïve OTSCC, we ended up with 239 patients. Those of the 239 patients who were primary

(28)

treatment-naive and treated in curative intent eligible for histopathological reassessment (n=150), formed the cohort for Paper III and Paper IV, as illustrated in the flow-chart, Figure 3.

Figure 3 Flow-chart outlining the number of patients and available histopathological samples in the NOROC cohort presented in Papers I-IV.

Rikshospitalet in Oslo is the national referral hospital. When a patient was located in EHRs at both the Rikshospitalet and another university hospital (registered twice), the case was

excluded from one of the hospital’s CRF in order to be registered as a single patient at a single hospital. Most cases were recorded in the CRF belonging to the hospital that completed the follow-up. The proportion of patients in each of the four different hospitals is shown in Figure 4.

643

^•ORAL CAVITY CANCER→

535 primary treatment- naïve squamous cell carcinomas (Paper I)

273

^•ORAL TONGUE SQUAMOUS CELL CARCINOMAS→146 eligible for HPV detection (Paper II)

239

•PRIMARY, TREATMENT- NAÏVE ORAL TONGUE SQUAMOUS CELL

CARCINOMA→150 available for histopathological reinvestigation (Papers III and IV)

(29)

Figure 4 Percentage and number of patients diagnosed with primary OCSCC at the four university hospitals in Norway, 2005-2009.

We retrieved clinical data such as gender, age, county, treating hospital, cigarette smoking, alcohol consumption, medication, comorbidity, ICD-10 diagnosis, treatment of surgery, radiation therapy and chemotherapy, follow-up, residual disease, recurrence of disease, second primary, and survival at fold-in date. See Appendix V for details.

3.2.2 HPV assessment

HPV analysis can be done by several methods: immunohistochemistry (IHC) for tumor suppressor protein (p)16, and HPV in situ hybridization (ISH) for both DNA and mRNA (35).

To be able to explore high-risk HPV in OTSCC, we used TMAs (85, 86). This is a method to analyze huge amounts of molecular analyses of, in this case, DNA, RNA, and p16, in so- called micro-matrices shown in Figure 5.

University Hospital of North Norway, 11% (n=56) St.Olavs University Hopsital, 12% (n=67)

Haukeland University Hospital, 17% (n=90)

Oslo University Hospital–Rikshopitalet, 60% (n=322)

(30)

Figure 5 Tissue microarrays of oral tongue squamous cell carcinoma: a) overview H&E- stained TMAs; b) example of one pan-keratin stained TMA block (EA1EA3). Photographs: T Søland.

TMAs were designed from formalin-fixed, paraffin-embedded tissue blocks from OTSCC samples. We used a fully automated TMA machine (Ventana). The p16 IHC, the HPV DNA ISH, and the HPV RNA ISH, were all performed on TMAs. In situ hybridization is used to map and order (in this case) DNA and RNA to localize a specific DNA or RNA in a section of tissue. For details regarding methodology, consult Paper II.

3.2.3 Extracting histopathological data

The histopathological re-evaluation was fully discussed among the NOROC collaborating pathologists before beginning the task and documenting into the CRF. The pathologists also had a more detailed written worklist corresponding to the pathology part in the CRF, shown in Appendix VI. The collaborating pathologists had workshops prior to histopathological assessment. The histological re-evaluation for the OTSCC was done on available

Hematoxylin and Eosin (H&E)-stained tissue sections and the pathologists were blinded for clinical outcome.

(31)

3.2.4 Depth of tumor invasion and tumor budding

Depth of tumor invasion or just depth of invasion is the depth the tumor invades into the tissue. The DOI was implemented for the T-status for OCC in the 8thedition of TNM in 2017 (22). This was also described by the International Consortium for Outcome Research in 2014 (24), and is also explained in Appendix VI. The pathologists re-evaluated the OTSCC and included the new DOI in the available H&E-slide sections.

Tumor budding is suggested as a new risk parameter for OTSCC (46). TB is a microscopic finding defined as the presence of isolated single tumor cells or small clusters of tumor cells, of four or fewer tumor cells at the invasive tumor front (47, 87), as shown in Figure 6. More details are explained in Paper III.

Figure 6 H&E-stained section of OTSCC with tumor budding; green arrowheads indicating buds<4, and black arrowheads indicating the tumor islands with>4 buds. Photographs: O Schreurs/T Søland.

(32)

3.2.5 Histopathological assessment of risk-patterns

Oral cavity SCC is for the most part described with the WHO differentiation of the whole tumor in the reports from the pathologists to the clinicians (21). Bryne and coworkers in 1998 and Brandwein-Gensler and coworkers in 2005, introduced assessment of histopathological risk-patterns in oral cancer (51, 52, 88). The additional histopathological information includes the WHO differentiation for worst pattern, nuclear polymorphism in the whole tumor, nuclear polymorphism at tumor front, keratinization of the whole tumor, keratinization in tumor front, lymphocyte infiltrate, perineural infiltration, and WPOI. A recent work by Steigen and

coworkers (2020) provides an overview of the variables (56), which is shown in Paper IV.

Details about grading are shown in Appendix IV. In Paper IV we also describe a less complex scoring model with fewer options, as previously shown to increase inter-rater and intra-rater agreement (56). All the histopathological variables assessed and presented in this thesis are listed below in Table 4.

Table 4. The various variables in oral tongue squamous cell carcinomas considered in this thesis.

Variable Score Paper

High-risk HPV Present or not II

Depth of tumor invasion (DOI) <5mm, 5-10mm, and >10mm III Tumor budding (TB) Low-high (2-tier) or

low-intermediate-high (3-tier) III

Tumor budding and depth of

invasion score Low-high (2-tier) or

low, intermediate, and high (3-tier) III

WHO differentiation, whole tumor Well-moderately-poor IV

WHO differentiation, worst pattern Well-moderately-poor IV

Degree of keratinization, whole

tumor Highly-moderately-minimal-no keratinization IV

Degree of keratinization, tumor

front Highly-moderately-minimal-no keratinization IV

Nuclear polymorphism, whole

tumor Little-moderately-abundant-extreme IV

Nuclear polymorphism, tumor front Little-moderately-abundant-extreme IV Perineural infiltration No nerves-at invasive front-in tumor center IV

Lymphocytic infiltrate Marked-moderate-little/none IV

Worst pattern of invasion (WPOI) Pushing-Infiltrating-small groups-marked-tumor

satellites (Type 1-5) IV

(33)

3.3 Data from the Cancer Registry of Norway

This study also uses data from the CRN. To be able to verify whether we had found the majority of patients diagnosed and treated for OCC, we contacted the CRN in October 2015 to obtain the total number of patients diagnosed with OCC (ICD-10, C02-C06) recorded in their registry between 2005–2009. The data presented in this thesis are those selected by the author from that official public document of the CRN. Data was provided according to current regulations; Legal Authority Cancer Registry § 3-4(1).

3.4 Data from Norwegian Cause of Death Registry

Cause of death was acquired from Norwegian Cause of Death Registry. In this manner we could calculate overall survival and disease-specific survival (89).

3.5 Statistical analyses

In this thesis, the following statistical methods have been performed by using the IBM Statistical Package for the Social Sciences (SPSS) version 25-26 for Windows (SPSS, Inc., Chicago, IL, USA). Descriptive data with range, mean, median, and frequencies, were used to characterize the cohort in the four included papers. Chi-square was used to describe

association among variables (90). Spearman bivariate correlation (2-tailed) including bootstrapping was used to identify correlation between variables (91). To avoid risk of bias from sparse-data, the variables with few data were excluded from these calculations (92). The Log-Rank (Mantel Cox) univariate survival analysis, giving Kaplan-Meier survival curves, was used to carry out a survival plot where all patients were censored at death or at follow-up after five years (93). When statistical variables were significant in univariate analysis, they were entered onto Cox regression multivariate analyses to assess their independent value as a prognostic factor of survival in the presence of other variables (90). For survival evaluation in multivariateanalyses, Cox regression allowed us to describe significance, hazard ratio (HR),

(34)

and 95% confidence interval (CI) after bootstrapping. Collinearity was evaluated with linear regression (90). All results were considered significant if p ≤0.05.

(35)

(36)

4 Summary of results

This is a summary of the papers in the present thesis. Paper I is based on the clinical characteristics of the oral cavity cancer patients in the NOROC study. Paper II is based on tissue microarrays assessment of high-risk HPV in oral tongue cancer. In Paper III and Paper IV, we focus on the different histopathological patterns in oral tongue squamous cell carcinomas and their association with survival outcomes.

4.1 Paper I

Characteristics and prognosis of primary treatment-naïve oral cavity squamous cell carcinoma in Norway, a descriptive retrospective study. Bjerkli IH, Jetlund O, Karevold G, Karlsdóttir Á, Jaatun E, Uhlin-Hansen L, Rikardsen OG, Hadler-Olsen E, Steigen SE.

PLoS One. 2020 Jan 16;15(1): e0227738. doi: 10.1371/journal.pone.0227738. eCollection 2020.

In Paper I, we outline a large and retrospective study of primary OCSCC from all four university hospitals in Norway, the NOROC study. We describe the incidence of the cancer at the different anatomical sites of the oral cavity, together with descriptive data such as gender, age, smoking, alcohol consumption, treatment of choice, follow-up, and five-year survival outcomes. The patients in the NOROC study form the basis of the other papers in this thesis.

All patients were diagnosed in the years 2005–2009, and validated against comparable cohorts.

• We identified 535 patients with primary treatment-naïve squamous cell carcinomas from the years 2005–2009.

• The male: female ratio was 1.2. Median age at diagnosis was 67 years, the range from 24 to 101 years. Men were eight years younger at diagnosis compared to women, median of 64 years and 72 years, respectively.

• Forty-five percent of the cases were oral tongue cancer.

• Age at time of diagnosis, tumor-status and node-status significantly influenced survival, but there was no gender difference in survival.

• The five-year disease-specific survival for the whole cohort was 52%. Patients treated with curative intent had a 62% five-year DSS.

(37)

4.2 Paper II

High-risk human papilloma virus was not detected in a Norwegian cohort of oral squamous cell carcinoma of the mobile tongue. Søland TM, Bjerkli IH, Georgsen JB, Schreurs O, Jebsen P, Laurvik H, Sapkota D. Manuscript submitted for publication.

In Paper II, we wanted to describe to what extent high-risk HPV was present in OTSCC.

Tissue microarrays of tumor tissue from 146 Norwegian OTSCC patients were assessed for presence of high-risk HPV. We compared different and independent approaches for HPV detection. We used DNA and RNA in situ hybridization assays and immunohistochemistry to evaluate the expression of HPV surrogate marker p16. We hypothesized that high-risk HPV infection is uncommon in OTSCC.

• We did not identify any transcriptional active HPV in these cases.

• None of the tumors were positive for either high-risk HPV DNA or for E6/E7 mRNA by in situ hybridization.

• Only two cases showed strong and uniform p16 in more than 70% of the cells (both cytoplasmic and nuclei staining).

• The main conclusion was that high-risk HPV is an unlikely causative factor in OTSCC.

4.3 Paper III

Tumor budding score predicts lymph node status in oral tongue squamous cell carcinoma and should be included in the pathology report. Bjerkli IH, Laurvik H, Nginamau ES, Søland TM, Costea D, Hov H, Uhlin-Hansen L, Hadler-Olsen E, Steigen SE.

Manuscript submitted for publication.

In Paper III, 150 cases of H&E-stained slide sections from OTSCC treated in curative intent were described. Survival outcome was compared with clinical and histopathological variables to evaluate their prognostic significance. We elucidated the prognostic value of the TNM 8th edition pT classification, including DOI, as well as lymph node status, TB, and a combined TB and DOI score, and whether these could supplement treatment decision.

(38)

• Tumors shifted to a higher T-status when classified according to the TNM 8th edition pT compared to the older pT. This indicates that DOI significantly complements T classification for prognostication.

• The age at diagnosis, TNM 8th edition pT classification, N-status, TB, DOI in a 3-tier category, and the combined TB and DOI scores, all significantly influenced prognosis in univariate analyses.

• A high TB score was associated with lymph node metastasis. For tumors with low TB score 22.5% had lymph node metastases, whereas for tumors with high TB score, 42.8% had lymph node metastases.

• The 8th edition T-status and TB in a 3-tier category were independent variables of five-year DSS in multivariate analyses.

• Tumor budding was associated with lymph node metastases, and can be used as a supplement to TNM classification in treatment decision for low-stage tumors.

4.4 Paper IV

A combined histo-score based on tumor differentiation and lymphocytic infiltrate is a robust prognostic marker for mobile tongue cancer. Bjerkli IH, Hadler-Olsen E,

Nginamau ES, Laurvik H, Søland TM, Costea D, Uhlin-Hansen L, Steigen SE. Manuscript submitted for publication.

In Paper IV, we determined whether tumor histopathological differentiation and other histopathologic high-risk assessments, when made less complex through fewer options, can help in predicting survival outcome for oral tongue cancer and thereby supplement treatment decisions.

• One-hundred and fifty H&E-stained sections of OTSCC treated in curative intent were available for re-investigation; 77 of these were low-stage tumors (T1-2, N0M0).

• Five-year disease-specific survival for the whole cohort was 65%, and for the low- stage tumors, five-year DSS was 83%.

• For the whole cohort, lymph node status and risk-patterns including differentiation of the whole tumor, perineural infiltration, and lymphocytic infiltrate, were found to be survival predictors.

(39)

• The WHO whole tumor differentiation correlated with survival outcome following the traditional grading, but also when made less complex, both for the whole cohort and for low-stage diseases.

• Lymphocytic infiltration can be scored with fewer options, to increase reproducibility (inter-rater and intra-rater agreeability), without losing prognostic value.

• A histo-score combining WHO differentiation and lymphocytic infiltration identified a group of low-stage tumors with a five-year DSS worse than the high-stage tumors, and this group of patients should be given special attention concerning treatment planning and follow-up.

(40)

5 Discussion

The general objective of this thesis has been to gain knowledge about cancer of the oral cavity from a cohort of strict OCSCC where the clinical and histopathological registration has been done in a structured manner. We present a large and well-controlled group of cases. As there may be differences in risk factors and clinical course of OCSCC from different oral sub-sites, we performed separate and more extensive analyses on the OTSCC, which comprised almost half of the cohort. Our retrospective study has not interfered with any choice of treatment or survival outcomes, insofar as the patients included had had their treatment 10-15 years ago.

This section will first address the ethical considerations of this research, emphasizing issues regarding the patient cohort. Then methodological and statistical considerations specific to this thesis will be addressed. The findings from the four papers will then be comprehensively discussed and related to other findings from the literature.

5.1 Ethical considerations

The Declaration of Helsinki was developed by the World Medical Association (WMA) in 1964, and has been updated several times (94). The declaration is a statement of ethical principles for medical research when it involves human subjects, and also covers human material and research data. The principle of informed consent is essential.

Originally this research was planned as a multicenter-study with two Norwegian university hospitals and one Finnish university hospital, but inasmuch as we found it difficult to collect clinical data in an unrelated foreign language, we shifted to a multicenter-study comprising the four university hospitals in Norway, naming it the NOROC study.

For the NOROC study, a passive consent procedure was utilized. The NOROC study and approach for consent was approved by the IRB of the REK Nord (Appendix I and Appendix II). Approval from one of the four regional IRBs gives validated approval for the other national health regions. Patients alive at the retrospective inclusion date received a written informal consent letter informing them about the study (Appendix III). The patients were

(41)

not opt-out were included in the analyses, except for three patients with written statements in the EHR about not participating in research; these were excluded at the outset.

This patient group is a vulnerable patient group having high mortality risk, with location of cancer in a part of the body that is important to vital functions and social life. Confidentiality of patient information is mandatory. Individual data such as information about burden of disease may have implications, for example, for the right to insurance; thus confidentiality is essential (95).

The study was planned retrospectively and approved in 2013/2014, five to ten years after the patients had had their cancer diagnosed and treated. The patients still alive were informed that they could withdraw from the NOROC study without concern.The letter was sent to patients when inclusion started between August 2015 and February 2017. We found the latest

addresses given in the EHR; addresses are updated once or twice a year. No letters were returned from either patients or postal services. Three patients contacted the principal

investigator by telephone to confirm they were agreeing to participate. One of them said, “do whatever you want, because this is very important research for our group of patients”, and another said, “you can do what you want at any time, and there is no need to inform me more about further research on my case”. My own experience with patients suggests that the majority of patients are very positive to research. None of the living patients contacted us to opt-out.As this was a retrospective study, many years after primary treatment, this study did not interfere with their treatment or outcome in any way.All data are kept anonymous, and the study register will be terminated when the time-period for the IRB approval closes, the study was given an extension in 2019 (Appendix IV).

If we had been obliged to have the patients to opt-in to the project, we could have been facing many patients not responding for several reasons, given their advanced age and perhaps not able to focus on an illness they survived many years ago. We would then only have had the clinical data and histopathological data of those not alive, and the analyses of five-year OS and DSS would not have been valid for all the patients with the cancer. With the consent approach we were given, we could gain insight into the patients who have survived, with both clinically and histopathological data, not only from those deceased. With our procedure, clinical data of all patients were available regardless of outcome or available tumor samples.

Prognostic indicators of survival for patients with oral cavity squamous cell carcinoma in Norway. Outcomes in a retrospective, multicenter cohort, with special focus on oral tongue squamous cell carcinoma, 2005-2009